Quantitative Detection Of Defect Area Of PVC Wedge Groove Using Infrared Thermal Wave

At present,the infrared thermal wave nondestructive technology has converted from the qualitative diagnosis and evaluation of defects into the quantitative detection of defects.However,due to the poor contrast of thermal images,the quantitative detection of defects is difficult.The quantitative detection of defect size is one of the difficulties,and how to select a thermal image that can show the defect feature better from the thermal image sequence is a primary problem to be solved urgently.The infrared thermal wave nondestructive testing of PVC plates with different depths of artificial wedge grooves by double side method were carried out.Select a frame that can display the image defect size from image sequences,use five kinds of the threshold image segmentation algorithms including: iterative threshold segmentation algorithm,maximum entropy segmentation algorithm,the minimum error method,Ostu method and minimum skewness segmentation method for image processing,and inquiry suitable algorithm for processing thermal image segmentation algorithm.In order to accurately extract the area of different depth of the defect size,the "divide and rule" strategy was put forward.The infrared image was divided into local image containing only one defect,and different burial depth defects were solved one by one.The comparison and analysis of the change of various data including threshold value,defect area,defect pixel and relative error of defect area with different algorithms in the "divide and rule" strategy before and after were carried in the interest of selecting the appropriate image processing method for extracting defects in infrared image.In order to quickly and accurately select a frame of thermal images that can reflect the size of the defect from the thermal images,this paper proposed maximum method of local mean square deviation method based on the definition of the sensitive area and analyzed the robustness of the method.The sensitive area and its size were randomly selected.100 independent repeat experiments were carried out by using the iterative threshold segmentation method and the Ostu method respectively,and the relative error ofdefect area was within 6% and 8% respectively.The results show that the "divide and rule" strategy is correct and effective for material detection with different buried depth defects.Iterative threshold segmentation algorithm and Ostu method are more suitable than other three algorithms for feature extraction of different buried depth defects.The maximum method of local mean square deviation method can be used to select a reasonable frame thermal image from the thermal image sequence for the defects of different depth so as to extract the defects accurately.The method is reliable and easy to operate,and it can lay the foundation for automatic processing of infrared NDT data.